On March 18--19, 1993, when Geotail was beyond 150 RE down the tail, for about 30 hours it frequently transited between magnetosheath-like plasma and lobe-like plasma. The transitions take the form of smooth variation by about a factor of 10 in both density and speed. Some of the transitions are distinctly asymmetric with a fast rise to sheath-like values and a slow decline to lobe-like values. We have suggested that these transitional regions are the plasma mantle, but this raises the question of why the mantle crossed Geotail many times in 30 hours. Three possible causes are flapping of the tail in response to solar wind flow changes (the windsock mechanism), intrinsic expansions and contractions of the tail boundary, perhaps in response to substorm phases (the breathing mechanism), and an IMF-squeezed elliptical tail cross section changing its orientation to follow the IMF (the wrenching mechanism). We test these possibilities here by examining simultaneous solar wind velocity and magnetic field data from IMP 8. For the cases studied, both the windsock and the breathing mechanisms appear to contribute to the motions that cause the transitions, whereas the wrenching mechanism seems less effective. Breathing dominates on a timescale of tens of minutes, and windsock dominates on a scale of hours. Since the windsock mechanism is unavoidable, the important finding here is that the breathing mechanism appears also to operate. We use the windsock variation to estimate the thickness of the mantle at Geotail's position and find that the density drops by a factor of 10 in 9 RE. We model the mantle as a one-dimensional slow-mode expansion fan and use the model to predict the change in plasma parameters that occurs from the outer edge of the mantle to its inner edge. A comparison of the predicted profile with the observed profile shows that the model simulates the observed changes reasonably well. ¿ American Geophysical Union 1996